Author(s): Lamas, Sandra ; Rodrigues, Nuno ; Santamaria-Echart, Arantzazu ; Palu, Igor ; Manhique, Jocyla ; Herrero, Baudilio ; Lopéz-Cortés, Isabel ; Pereira, José Alberto ; Peres, António M.
Date: 2025
Persistent ID: http://hdl.handle.net/10198/34513
Origin: Biblioteca Digital do IPB
Subject(s): Dry fruits; Cultivar traceability; Principal Component Analysis; Linear Discriminant Analysis
Three almond cultivars (Lauranne, Marinada, and Vairo) were studied, considering morphological parameters that showed statistical cultivar-dependence but not enabled accurate cultivar recognition. Alternatively, attenuated total reflectance Fourier transform infrared spectroscopy (ATR-FTIR) was applied to whole endocarp, kernel, and ground almonds. Both transmittance spectra and respective derivatives (3400–2800 and 1900–600 cm of the three matrices were used to establish multivariate linear discriminant models, based on subsets of selected wavenumbers (6–33), allowing 90.5–99.7% of correct cultivar classification for repeated K-fold cross-validation. Ground almonds yielded the best results regardless of spectrum pre-treatment. While epicarp analysis offers less invasiveness, the use of raw transmittance spectra of ground almonds resulted into the most practical approach due to the need of fewer independent variables (less complex models), proving effective for cultivar identification via ATR-FTIR -chemometric tools. Overall, the findings point out that ATR-FTIR is a reliable tool for almond cultivar traceability.
